Yet another Federer/Nadal thread (common opponent analysis)

73west

Semi-Pro
Starting a thread because this could be responsive to any one of a few other active threads. I wanted to put forward a statistical analysis that I've been building for a while. The background is this: I had a conversation with a friend about Federer vs Nadal H2H (this was before Fed's run in 2017) and whether it indicates that Nadal is the better player, or that it's a bad matchup for Federer, or just that it's a statistical quirk, etc. I posited that *if* it is just a matchup issue, that Federer specifically has a style of play or mental hangup with Nadal, then there is an easy test. If instead of Federer vs Nadal, you look at (Federer vs Player X) vs (Nadal vs Player X) and keep changing X, Federer should win most of those. If he does, it suggests it is a matchup issue or statistical quirk. If Nadal wins most of those, on top of H2H, it suggests he is the better player.

So I picked a sample of Player Xs. I took every player who fit these criteria
1. They have finished YE Top 5 at least once in their career (so I'm only looking at elite opponents)
2. They have played both Federer and Nadal at least 5 times each (so I have a meaningful sample)

There are 15 such players as of today.
Djokovic, Murray, Wawrinka, Roddick, Del Potro, Hewitt, Ferrero, Moya, Ferrer, Davydenko, Soderling, Nishikori, Raonic, Dimitrov and Blake.

For each, I ran 3 numbers/stats
Stat 1. Career H2H, and whether 1 player (Federer or Nadal) has an edge. I give one of them an edge over the other if their winning %age is better by enough to generate at least 0.5 extra wins given the number of matches played (I use the lower matches played total, between Fed/Nadal). So, to give 2 examples:
Federer is 22-23 vs Djokovic, or 48.9%. Nadal is 24-26, or 48.0%.
(48.9%-48.0%) * (45) = 0.4 extra wins for Federer. That is less than 0.5, so I call that a "draw". They basically have the same record against Djokovic.
Federer is 14-11 against Murray, or 56%. Nadal is 17-7, or 71%.
(71%-56%) * (24) = 3.6. So Nadal has an edge of 3.6 games over Federer when it comes to the Murray H2H

Stat 2. H2H when the opponent was at his peak. To define peak, I looked at each opponent's year end rank, picked a meaningful cutoff point (such as "the 4 years where he finished in the top 5") and took the H2H only for those years. For example, I defined Peak Djokovic as the 5 year period where he had 4 YE #1 finishes, so I took the H2H over those 5 years. I call that H2H against Peak Djokovic. Then I did the same calc as above. Example
Federer is 9-16 (36%) against Peak Djokovic. Nadal is 7-16 (30%).
(36% - 20%) * (23) = 1.3, so Federer has an edge of 1.3 games over nadal when it comes to Peak Djokovic H2H
Note: for players like Hewitt, whose peak preceded both Fed and Nadal, I am not considering the years where he did not play Federer/Nadal in my definition of peak.

Stat 3. How much of the H2H was against the opponent's peak.
This is a simple one, it's just taking the # of matches from (2) divided by the number of matches from (1). Example, Federer has played Djokovic 45 times, and 25 were against Peak Djokovic. 25/45 = 56%. Nadal has played Djokovic 50 times, and 23 were against peak Djokovic. 17/50 = 46%. So Federer's Djokovic H2H is slightly more weighted to Peak Djokovic than Nadal's is.

I've included the chart, but analysis will be in a separate post
(http://iblogforcookies.com/peak.PNG ... in case the embed is too small to read)

peak.PNG
 

73west

Semi-Pro
Analysis

This seems to pretty conclusively suggest that Federer *does* consistently do better against the elite common opponents.
His composite H2H is 76%, compared to 68% for Nadal, a pretty sizeable edge.
Of the 15 comparisons, Federer is ahead in 9, Nadal is ahead in 2 and the other 4 are "draws"
If you just include peak years for opponnets, Federer's composite is 70%, compared to 51%, a huge edge (Nadal being just above 50% is not great, for this comparison), and Federer wins 9 comparisons, to Nadal's 0 (the rest being draws)

And overal, they are just as likely to draw opponents' peaks (41% for Federer to 37% for Nadal) so it's not biased by facing off-peak opponents.
 
Last edited:

73west

Semi-Pro
In short:

Fedr GOAT

Well, it's only Fed vs Nadal.
I am working on Fed vs Djoko, Nadal vs Djoko in a similar format, but it's not ready.
The interesting thing is so far, it looks like Fed vs Djoko is a much closer comparison than Fed vs Nadal, but I haven't broken down the peak / non-peak part of it yet.
 

mike danny

Bionic Poster
@73west, Fed played peak Djokovic 26 times, not 25. Maybe it would influence the overall percentage a bit.

Look at 2011: Fed defeated peak Djokovic that year, while Nadal failed to do it. A 20% win percentage still vastly better than 0% winning percentage.
 

73west

Semi-Pro
@73west, Fed played peak Djokovic 26 times, not 25. Maybe it would influence the overall percentage a bit.

Look at 2011: Fed defeated peak Djokovic that year, while Nadal failed to do it. A 20% win percentage still vastly better than 0% winning percentage.

I have him down for 5 matches in 2011, 5 in 2012, 2 in 2013, 5 in 2014, 8 in 2015, so 25. There's a 26th which I didn't count, because it was a walkover.
I know peak definitiosn are extremely subjective, but I just wanted to do something simple and easy to calculate.

And hear you on 2011. Also 2015, where he actually beat him 3 times. At the level Djokovic was at, 3-5 is an impressive result.
 
Very interesting analysis. I also wonder how restricting some of this to "big" matches such as slams would play out. As an example, nadal lost quite a few matches to peak Djokovic, even on clay, but did much better in the slams.
 
I have him down for 5 matches in 2011, 5 in 2012, 2 in 2013, 5 in 2014, 8 in 2015, so 25. There's a 26th which I didn't count, because it was a walkover.
I know peak definitiosn are extremely subjective, but I just wanted to do something simple and easy to calculate.

And hear you on 2011. Also 2015, where he actually beat him 3 times. At the level Djokovic was at, 3-5 is an impressive result.
2016 AO SF.
 
D

Deleted member 757377

Guest
The top ranked opponent Federer beat at Australian Open is n.3 Ferrero.

Slam finals won with a 22-29 year old 3+ slam champion: 1
 

mike danny

Bionic Poster
I have him down for 5 matches in 2011, 5 in 2012, 2 in 2013, 5 in 2014, 8 in 2015, so 25. There's a 26th which I didn't count, because it was a walkover.
I know peak definitiosn are extremely subjective, but I just wanted to do something simple and easy to calculate.

And hear you on 2011. Also 2015, where he actually beat him 3 times. At the level Djokovic was at, 3-5 is an impressive result.
I was thinking about 2015, but Nadal himself was in poor form so I didn't take it into account.

The 26th match was in 2016 at the AO. That was definitely peak Djokovic. But I see you only considered YE#1 seasons so my bad.
 

73west

Semi-Pro
I was thinking about 2015, but Nadal himself was in poor form so I didn't take it into account.

The 26th match was in 2016 at the AO. That was definitely peak Djokovic. But I see you only considered YE#1 seasons so my bad.

Djokovic was clearly still a peak player, but when I do statistical analysis I try really hard to use hard cutoffs so that I don't get dragged into problems with subjectivity.
 

Zeref

Professional
The top ranked opponent Federer beat at Australian Open is n.3 Ferrero.

Slam finals won with a 22-29 year old 3+ slam champion: 1
I don't blame you...this actual representation of stats must have gone over your head.
 

73west

Semi-Pro
@73west :

Pretty sure Fed is undefeated vs Moya and lost only once to Blake.

You're right. The tail end of the Peak H2H column (Ferrero, Blake, Moya) had Nadal's and Federer's records reversed. I've corrected it. I also edited the Peak "advantage" column, and put in a 3 match minimum (if an opponent hasn't had at least 3 peak level matches against each of Fed/Nadal, then the comparison is deemed a draw).

Thanks!
 
Last edited:
Top